Billiards and Tilts Analysis Notebook

0.1 Load all of the throw data

#load omnibus dataframe
omnibus_df <- read_delim("../data/processed/omnibus/omnibus_raw.csv", 
                            delim = ",", 
                            col_types = cols(.default = col_double(), 
                                             type = col_factor(),
                                             ppid = col_factor(),
                                             exp_label = col_factor(),
                                             experiment = col_factor(),
                                             hand = col_factor(),
                                             camera_tilt = col_factor(),
                                             surface_tilt = col_factor(),
                                             target = col_factor(),
                                             test_type = col_factor(),
                                             prior_anim = col_factor(),
                                             baseline_block = col_factor(),
                                             task_type = col_factor(),
                                             surface = col_factor(),
                                             anim_type = col_factor()))

original_exps <- c("rot15_cued_tilt", "rot15_uncued", "tilt_uncued_rot", 
                   "tilt_uncued_norot", "tilt_cued_rot", "tilt_cued_norot")

1 Visualizing data (univariate)

test_ppt <- 3

test_df <- omnibus_df %>% filter(ppid == test_ppt)


trial <- 250
trial_df <- filter(test_df, trial_num == trial)

x <- trial_df$flick_velocity_x
y <- trial_df$flick_velocity_y
z <- trial_df$flick_velocity_z

x2 <- trial_df$flick_direction_x
y2 <- trial_df$flick_direction_y
z2 <- trial_df$flick_direction_z

# plot both
plot_ly(x = c(0, x), y = c(0, y), z = c(0, z), type = "scatter3d", mode = "lines") %>%
  add_trace(x = c(0, x2), y = c(0, y2), z = c(0, z2), type = "scatter3d", mode = "lines") %>%
  layout(scene = list(xaxis = list(title = "x"),
                      yaxis = list(title = "y"),
                      zaxis = list(title = "z")))

# note: this is a rotated trial

# plot distribution of error_size
p <- ggplot(omnibus_df, aes(x = error_size)) +
  geom_histogram(binwidth = .5) +
  theme_minimal() +
  theme(text = element_text(size = 11)) +
  labs(x = "Error Size (cm)", y = "Count")

p

2 Group

2.1 All throws

# try just animate_surface exp
animate_surface_trial_summary_df <- omnibus_df %>%
  filter(exp_label == "animate_surface") %>%
  group_by(prior_anim, block_num, trial_num_in_block, trial_num) %>%
  summarise(
    mean_deviation = mean(throw_deviation),
    ci_deviation = vector_confint(throw_deviation)
  )

`summarise()` has grouped output by 'prior_anim', 'block_num', 'trial_num_in_block'. You can override using the `.groups` argument.

data <- animate_surface_trial_summary_df

# set up plot
p <- data %>%
  ggplot(
    aes(
      x = trial_num, y = mean_deviation,
      ymin = mean_deviation - ci_deviation,
      ymax = mean_deviation + ci_deviation
    )
  ) +
  theme_classic() +
  theme(legend.position = "none") +
  labs(
    x = "Trial Number",
    y = "Throw Angle (°)"
  )

# add horizontal lines
p <- p +
  geom_hline(
    yintercept = c(0, -30), linewidth = 0.4,
    colour = "#CCCCCC", linetype = "solid"
  ) +
  geom_hline(
    yintercept = c(-15), linewidth = 0.4,
    colour = "#CCCCCC", linetype = "dashed"
  )

# p <- p +
#   scale_y_continuous(
#     limits = c(-10, 35),
#     breaks = c(0, 15, 30),
#     labels = c(0, 15, 30)
#   ) +
#   scale_x_continuous(
#     limits = c(0, 180),
#     breaks = c(0, 60, 120, 180),
#     labels = c(0, 60, 120, 180)
#   )

# set font size to 11
p <- p +
  theme(text = element_text(size = 11))

# repeat for prior_anim == "half", "full" and "wait"
for (unique_prior_anim in unique(data$prior_anim)) {
  # get the data for this block
  to_plot_data <- filter(data, prior_anim == unique_prior_anim)
  # loop through the unique blocks in to_plot_data
  for (block in unique(to_plot_data$block_num)) {
    # get the data for this block
    block_data <- filter(to_plot_data, block_num == block)
    # add the data, use the pallete_list to get the colour
    p <- p + geom_ribbon(
      data = block_data,
      aes(fill = prior_anim),
      colour = NA, alpha = 0.3
      ) + geom_line(
      data = block_data,
      aes(colour = prior_anim))
    
  }
}


# # save
# if (save_plots) {
#   ggsave(
#   p,
#   filename = "../plots/paper_figs/sr_30_training.pdf", device = "pdf",
#   height = 4, width = 6
#   )
#   }

ggplotly(p)

# p

# rest of the exps
original_exps_trial_summary_df <- omnibus_df %>%
  filter(exp_label == "original_exps") %>%
  group_by(experiment, trial_num) %>%
  summarise(
    mean_deviation = mean(throw_deviation),
    ci_deviation = vector_confint(throw_deviation)
  )

data <- original_exps_trial_summary_df

# set up plot
p <- data %>%
  ggplot(
    aes(
      x = trial_num, y = mean_deviation, colour = experiment
    )
  ) +
  theme_classic() +
  theme(legend.position = "none") +
  labs(
    x = "Trial Number",
    y = "Throw Angle (°)"
  )

# add horizontal lines
p <- p +
  geom_hline(
    yintercept = c(0, -30), linewidth = 0.4,
    colour = "#CCCCCC", linetype = "solid"
  ) +
  geom_hline(
    yintercept = c(-15), linewidth = 0.4,
    colour = "#CCCCCC", linetype = "dashed"
  )

# p <- p +
#   scale_y_continuous(
#     limits = c(-10, 35),
#     breaks = c(0, 15, 30),
#     labels = c(0, 15, 30)
#   ) +
#   scale_x_continuous(
#     limits = c(0, 180),
#     breaks = c(0, 60, 120, 180),
#     labels = c(0, 60, 120, 180)
#   )

# set font size to 11
p <- p +
  theme(text = element_text(size = 11))

# add confidence intervals
p <- p + geom_ribbon(
  aes(
    ymin = mean_deviation - ci_deviation,
    ymax = mean_deviation + ci_deviation,
    fill = experiment
  ), colour = NA, alpha = 0.3
  )


# add data
p <- p + geom_line()


# # save
# if (save_plots) {
#   ggsave(
#   p,
#   filename = "../plots/paper_figs/sr_30_training.pdf", device = "pdf",
#   height = 4, width = 6
#   )
#   }

ggplotly(p)

# p

ggplotly(plot_original_learning_curve(omnibus_path))

visible vs non-visible tilt doesn’t affect the 15-degree rotation condition. But affects all other conditions. So 15-degree rotation

ggplotly(plot_original_rebound(omnibus_path))

Plot all the trials and see if there are group effects.

all_throw_summary <- omnibus_df %>%
  group_by(experiment, trial_num) %>%
  summarise(mean = mean(error_size, na.rm = TRUE),
            sd = sd(error_size, na.rm = TRUE), 
            ci = vector_confint(error_size),
            n = n(), .groups = "drop")

p <- omnibus_df %>%
  ggplot(aes(x = trial_num, y = error_size)) +
  # geom_point(alpha = 0.04, aes(colour = experiment)) + 
  geom_ribbon(data = all_throw_summary, 
                 aes(y = mean, ymin = mean-ci,
                     ymax = mean+ci, fill = experiment), 
              alpha = 0.3) +
  geom_line(data = all_throw_summary, 
            aes(y = mean, colour = experiment)) +
  theme_minimal() +
  # scale_x_continuous(limits = c(205, 374)) +
  NULL

# ggplotly(p)
p
# save plot
# ggsave(p, height = 9, width = 16, device = "svg", filename = "data/figs/all_throws.svg")

2.2 Success manifolds

2.2.1 Without any tilts

# ggplotly(plot_success_manifold_no_tilt())
plot_success_manifold_no_tilt()

2.2.2 With tilt present

ggplotly(plot_success_manifold_tilt())